Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7fc427ce97b8>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7fc427bdcf28>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    input_images = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name="real_image_input")
    z_data = tf.placeholder(tf.float32, (None, z_dim), name="z_input")
    learning_rate = tf.placeholder(tf.float32, None, name="learning_rate")

    return input_images, z_data, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def leakyrelu(x, alpha=0.2):
    """
    Calculates a leakyrelu function
    :param x: input tensor
    :param alpha: defines the leakyness, values below 0 will get multiplied with this
    :return: output tensor
    """
    return tf.maximum(alpha * x, x)

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    
    def convo2d(inputs,  filters, batch_norm=True):
        outputs = tf.layers.conv2d(inputs, filters, 5, 2, padding="same")
        if batch_norm:
            outputs = tf.layers.batch_normalization(outputs, training=True)
        return leakyrelu(outputs)
        
    
    with tf.variable_scope("discriminator", reuse=reuse):
        # input 28*28*1 for mnist, 28*28*3 for celebA
        h1 = convo2d(images, 64, batch_norm=False)
        # 14*14*64
        h2 = convo2d(h1, 128)
        # 7*7*128
        h3 = convo2d(h2, 256)
        # 4*4*256
        
        flat = tf.reshape(h3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    
    with tf.variable_scope("generator", reuse=not is_train):
        g1 = tf.layers.dense(z, 7*7*512)
        g1 = tf.reshape(g1, (-1, 7, 7, 512))
        g1 = tf.layers.batch_normalization(g1, training=is_train)
        g1 = leakyrelu(g1)
        # 7*7*512
        
        g2 = tf.layers.conv2d_transpose(g1, 256, 5, strides=2, padding="same", activation=None)
        g2 = tf.layers.batch_normalization(g2, training=is_train)
        g2 = leakyrelu(g2)
        # 14*14*256
        
        g3 = tf.layers.conv2d_transpose(g2, 128, 5, strides=2, padding="same", activation=None)
        g3 = tf.layers.batch_normalization(g3, training=is_train)
        g3 = leakyrelu(g3)
        # 28*28*128
        
        # output layer
        logits = tf.layers.conv2d_transpose(g3, out_channel_dim, 5, strides=1, padding="same", activation=None)
        # 28*28*out_channel_dim now
        
        output = tf.tanh(logits)
    return output


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    
    smooth = 0.1
    
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_real, 
            labels=tf.ones_like(d_model_real) * (1- smooth))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_fake, 
            labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(
            logits=d_logits_fake, 
            labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    
    # Get weights and bias to update
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    
    _, image_width, image_height, image_channels = data_shape
    
    input_real, input_z, lr = model_inputs(image_width, image_height, image_channels, z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, image_channels)
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    samples, losses = [], []
    steps = 0
    
    saver = tf.train.Saver()
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                steps += 1
                
                # So the values of the MNIST and CelebA dataset are in a range of -0.5 to 0.5
                batch_images *= 2
                
                # Sample ramdom noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                # Run optimizers
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z})
                _ = sess.run(g_opt, feed_dict={input_z: batch_z, input_real: batch_images})
                
                if steps % 10 == 0:
                    # At the end of each epoch, get the losses and print them out
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})
                
                    print("Epoch {}/{}...".format(epoch_i+1, epochs),
                            "Generator Loss: {:.4f}".format(train_loss_g),
                            "Discriminator Loss: {:.4f}...".format(train_loss_d))
                    # Save losses to view after training
                    losses.append((train_loss_d, train_loss_g))
                
                if steps % 100 == 0:
                    show_generator_output(sess, 16, input_z, image_channels, data_image_mode)
                    
        saver.save(sess, "./checkpoints/generator.ckpt")
        show_generator_output(sess, 16, input_z, image_channels, data_image_mode)
    return losses

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 64
z_dim = 100
learning_rate = 0.002
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Generator Loss: 8.4189 Discriminator Loss: 0.0002...
Epoch 1/2... Generator Loss: 34.0672 Discriminator Loss: 26.7649...
Epoch 1/2... Generator Loss: 1.8874 Discriminator Loss: 0.7783...
Epoch 1/2... Generator Loss: 3.0788 Discriminator Loss: 0.1050...
Epoch 1/2... Generator Loss: 13.2902 Discriminator Loss: 0.0000...
Epoch 1/2... Generator Loss: 0.0221 Discriminator Loss: 5.6617...
Epoch 1/2... Generator Loss: 0.0455 Discriminator Loss: 3.4776...
Epoch 1/2... Generator Loss: 0.5191 Discriminator Loss: 1.5037...
Epoch 1/2... Generator Loss: 2.2375 Discriminator Loss: 1.0835...
Epoch 1/2... Generator Loss: 2.5368 Discriminator Loss: 2.2419...

Epoch 1/2... Generator Loss: 2.3626 Discriminator Loss: 1.3683...
Epoch 1/2... Generator Loss: 0.6348 Discriminator Loss: 1.2755...
Epoch 1/2... Generator Loss: 1.4086 Discriminator Loss: 1.3453...
Epoch 1/2... Generator Loss: 1.5408 Discriminator Loss: 1.3265...
Epoch 1/2... Generator Loss: 0.3233 Discriminator Loss: 1.8487...
Epoch 1/2... Generator Loss: 0.3974 Discriminator Loss: 1.4304...
Epoch 1/2... Generator Loss: 0.2931 Discriminator Loss: 1.8154...
Epoch 1/2... Generator Loss: 0.3246 Discriminator Loss: 1.7533...
Epoch 1/2... Generator Loss: 0.1993 Discriminator Loss: 2.3122...
Epoch 1/2... Generator Loss: 1.0487 Discriminator Loss: 0.8973...

Epoch 1/2... Generator Loss: 0.6349 Discriminator Loss: 1.2397...
Epoch 1/2... Generator Loss: 0.7515 Discriminator Loss: 1.3567...
Epoch 1/2... Generator Loss: 1.5964 Discriminator Loss: 0.9120...
Epoch 1/2... Generator Loss: 1.0198 Discriminator Loss: 1.3847...
Epoch 1/2... Generator Loss: 1.2110 Discriminator Loss: 1.4368...
Epoch 1/2... Generator Loss: 2.0029 Discriminator Loss: 1.7390...
Epoch 1/2... Generator Loss: 0.9399 Discriminator Loss: 0.9973...
Epoch 1/2... Generator Loss: 0.9428 Discriminator Loss: 1.2365...
Epoch 1/2... Generator Loss: 0.8569 Discriminator Loss: 1.4328...
Epoch 1/2... Generator Loss: 1.1449 Discriminator Loss: 1.1145...

Epoch 1/2... Generator Loss: 1.3425 Discriminator Loss: 1.2889...
Epoch 1/2... Generator Loss: 1.3215 Discriminator Loss: 0.7460...
Epoch 1/2... Generator Loss: 2.7454 Discriminator Loss: 1.7720...
Epoch 1/2... Generator Loss: 0.7479 Discriminator Loss: 1.1274...
Epoch 1/2... Generator Loss: 2.1786 Discriminator Loss: 1.4116...
Epoch 1/2... Generator Loss: 1.0023 Discriminator Loss: 1.1273...
Epoch 1/2... Generator Loss: 1.0994 Discriminator Loss: 0.7767...
Epoch 1/2... Generator Loss: 0.7258 Discriminator Loss: 0.9810...
Epoch 1/2... Generator Loss: 0.4012 Discriminator Loss: 1.4105...
Epoch 1/2... Generator Loss: 1.9732 Discriminator Loss: 1.2703...

Epoch 1/2... Generator Loss: 0.9620 Discriminator Loss: 1.1219...
Epoch 1/2... Generator Loss: 1.3158 Discriminator Loss: 1.4413...
Epoch 1/2... Generator Loss: 0.6028 Discriminator Loss: 1.3138...
Epoch 1/2... Generator Loss: 0.6654 Discriminator Loss: 1.2422...
Epoch 1/2... Generator Loss: 1.1996 Discriminator Loss: 1.1975...
Epoch 1/2... Generator Loss: 1.5394 Discriminator Loss: 1.5190...
Epoch 1/2... Generator Loss: 1.2292 Discriminator Loss: 1.2181...
Epoch 1/2... Generator Loss: 0.4387 Discriminator Loss: 1.4722...
Epoch 1/2... Generator Loss: 0.3956 Discriminator Loss: 1.4617...
Epoch 1/2... Generator Loss: 0.7161 Discriminator Loss: 1.0738...

Epoch 1/2... Generator Loss: 0.4437 Discriminator Loss: 1.3610...
Epoch 1/2... Generator Loss: 1.7087 Discriminator Loss: 1.1750...
Epoch 1/2... Generator Loss: 1.8764 Discriminator Loss: 1.7596...
Epoch 1/2... Generator Loss: 0.2951 Discriminator Loss: 1.6321...
Epoch 1/2... Generator Loss: 0.3848 Discriminator Loss: 1.3107...
Epoch 1/2... Generator Loss: 0.6061 Discriminator Loss: 1.1342...
Epoch 1/2... Generator Loss: 0.3348 Discriminator Loss: 1.5231...
Epoch 1/2... Generator Loss: 0.3262 Discriminator Loss: 1.5860...
Epoch 1/2... Generator Loss: 0.4169 Discriminator Loss: 1.4366...
Epoch 1/2... Generator Loss: 1.4419 Discriminator Loss: 1.3192...

Epoch 1/2... Generator Loss: 1.5858 Discriminator Loss: 1.2761...
Epoch 1/2... Generator Loss: 0.9813 Discriminator Loss: 1.2583...
Epoch 1/2... Generator Loss: 0.8364 Discriminator Loss: 1.2879...
Epoch 1/2... Generator Loss: 0.5970 Discriminator Loss: 1.4720...
Epoch 1/2... Generator Loss: 0.3678 Discriminator Loss: 1.4858...
Epoch 1/2... Generator Loss: 0.3523 Discriminator Loss: 1.5320...
Epoch 1/2... Generator Loss: 0.6846 Discriminator Loss: 1.2270...
Epoch 1/2... Generator Loss: 0.3327 Discriminator Loss: 1.5357...
Epoch 1/2... Generator Loss: 1.3103 Discriminator Loss: 1.3033...
Epoch 1/2... Generator Loss: 1.3071 Discriminator Loss: 1.0294...

Epoch 1/2... Generator Loss: 1.7543 Discriminator Loss: 1.5483...
Epoch 1/2... Generator Loss: 0.4061 Discriminator Loss: 1.4108...
Epoch 1/2... Generator Loss: 0.4230 Discriminator Loss: 1.3750...
Epoch 1/2... Generator Loss: 0.3760 Discriminator Loss: 1.6958...
Epoch 1/2... Generator Loss: 0.5216 Discriminator Loss: 1.2816...
Epoch 1/2... Generator Loss: 0.2337 Discriminator Loss: 1.8031...
Epoch 1/2... Generator Loss: 0.4099 Discriminator Loss: 1.3740...
Epoch 1/2... Generator Loss: 0.4379 Discriminator Loss: 1.4788...
Epoch 1/2... Generator Loss: 0.8225 Discriminator Loss: 1.1815...
Epoch 1/2... Generator Loss: 0.5213 Discriminator Loss: 1.2721...

Epoch 1/2... Generator Loss: 0.8042 Discriminator Loss: 1.0850...
Epoch 1/2... Generator Loss: 0.2646 Discriminator Loss: 1.7187...
Epoch 1/2... Generator Loss: 0.8613 Discriminator Loss: 1.0886...
Epoch 1/2... Generator Loss: 0.4625 Discriminator Loss: 1.2435...
Epoch 1/2... Generator Loss: 0.3742 Discriminator Loss: 1.4663...
Epoch 1/2... Generator Loss: 0.2949 Discriminator Loss: 1.6960...
Epoch 1/2... Generator Loss: 0.6771 Discriminator Loss: 1.2003...
Epoch 1/2... Generator Loss: 1.2135 Discriminator Loss: 1.2294...
Epoch 1/2... Generator Loss: 1.3250 Discriminator Loss: 1.2524...
Epoch 1/2... Generator Loss: 0.3821 Discriminator Loss: 1.4537...

Epoch 1/2... Generator Loss: 0.3188 Discriminator Loss: 1.6171...
Epoch 1/2... Generator Loss: 0.5766 Discriminator Loss: 1.2858...
Epoch 1/2... Generator Loss: 0.3240 Discriminator Loss: 1.5558...
Epoch 2/2... Generator Loss: 0.3743 Discriminator Loss: 1.5357...
Epoch 2/2... Generator Loss: 0.3206 Discriminator Loss: 1.6309...
Epoch 2/2... Generator Loss: 0.8788 Discriminator Loss: 1.1338...
Epoch 2/2... Generator Loss: 1.6242 Discriminator Loss: 1.3935...
Epoch 2/2... Generator Loss: 0.6451 Discriminator Loss: 1.2038...
Epoch 2/2... Generator Loss: 0.3424 Discriminator Loss: 1.5902...
Epoch 2/2... Generator Loss: 0.5576 Discriminator Loss: 1.2945...

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 64
z_dim = 100
learning_rate = 0.0002
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 10

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/10... Generator Loss: 0.0905 Discriminator Loss: 2.8175...
Epoch 1/10... Generator Loss: 0.0754 Discriminator Loss: 3.1378...
Epoch 1/10... Generator Loss: 0.0538 Discriminator Loss: 3.1585...
Epoch 1/10... Generator Loss: 0.0956 Discriminator Loss: 2.5719...
Epoch 1/10... Generator Loss: 0.3951 Discriminator Loss: 1.4971...
Epoch 1/10... Generator Loss: 0.5764 Discriminator Loss: 1.0111...
Epoch 1/10... Generator Loss: 1.0342 Discriminator Loss: 0.6930...
Epoch 1/10... Generator Loss: 0.1771 Discriminator Loss: 2.0053...
Epoch 1/10... Generator Loss: 0.5298 Discriminator Loss: 1.1509...
Epoch 1/10... Generator Loss: 0.7096 Discriminator Loss: 0.9096...

Epoch 1/10... Generator Loss: 0.0633 Discriminator Loss: 2.9477...
Epoch 1/10... Generator Loss: 0.2847 Discriminator Loss: 1.6244...
Epoch 1/10... Generator Loss: 0.2065 Discriminator Loss: 1.9470...
Epoch 1/10... Generator Loss: 0.4071 Discriminator Loss: 1.3801...
Epoch 1/10... Generator Loss: 0.1130 Discriminator Loss: 2.4618...
Epoch 1/10... Generator Loss: 0.2243 Discriminator Loss: 1.8234...
Epoch 1/10... Generator Loss: 0.6158 Discriminator Loss: 1.0438...
Epoch 1/10... Generator Loss: 3.1200 Discriminator Loss: 0.6438...
Epoch 1/10... Generator Loss: 2.5220 Discriminator Loss: 1.2450...
Epoch 1/10... Generator Loss: 0.1643 Discriminator Loss: 2.3212...

Epoch 1/10... Generator Loss: 3.2471 Discriminator Loss: 0.5123...
Epoch 1/10... Generator Loss: 1.9086 Discriminator Loss: 0.7507...
Epoch 1/10... Generator Loss: 0.0821 Discriminator Loss: 2.7828...
Epoch 1/10... Generator Loss: 0.3585 Discriminator Loss: 1.4677...
Epoch 1/10... Generator Loss: 0.2076 Discriminator Loss: 1.8545...
Epoch 1/10... Generator Loss: 0.4370 Discriminator Loss: 1.2541...
Epoch 1/10... Generator Loss: 0.2735 Discriminator Loss: 1.7157...
Epoch 1/10... Generator Loss: 0.2229 Discriminator Loss: 1.8847...
Epoch 1/10... Generator Loss: 1.8142 Discriminator Loss: 0.8588...
Epoch 1/10... Generator Loss: 1.6968 Discriminator Loss: 1.0322...

Epoch 1/10... Generator Loss: 1.6631 Discriminator Loss: 2.0886...
Epoch 1/10... Generator Loss: 3.9167 Discriminator Loss: 0.6981...
Epoch 1/10... Generator Loss: 0.5995 Discriminator Loss: 1.0116...
Epoch 1/10... Generator Loss: 0.2324 Discriminator Loss: 1.7714...
Epoch 1/10... Generator Loss: 0.2685 Discriminator Loss: 1.6828...
Epoch 1/10... Generator Loss: 0.5887 Discriminator Loss: 0.9701...
Epoch 1/10... Generator Loss: 2.9641 Discriminator Loss: 0.6527...
Epoch 1/10... Generator Loss: 2.3673 Discriminator Loss: 0.2858...
Epoch 1/10... Generator Loss: 0.2284 Discriminator Loss: 1.7888...
Epoch 1/10... Generator Loss: 3.3070 Discriminator Loss: 0.5689...

Epoch 1/10... Generator Loss: 0.1943 Discriminator Loss: 1.9411...
Epoch 1/10... Generator Loss: 0.6317 Discriminator Loss: 1.3583...
Epoch 1/10... Generator Loss: 4.7766 Discriminator Loss: 1.1457...
Epoch 1/10... Generator Loss: 3.2757 Discriminator Loss: 0.2267...
Epoch 1/10... Generator Loss: 3.8751 Discriminator Loss: 0.7483...
Epoch 1/10... Generator Loss: 1.6657 Discriminator Loss: 0.4461...
Epoch 1/10... Generator Loss: 2.5501 Discriminator Loss: 3.1606...
Epoch 1/10... Generator Loss: 1.7617 Discriminator Loss: 0.3993...
Epoch 1/10... Generator Loss: 1.6335 Discriminator Loss: 0.4663...
Epoch 1/10... Generator Loss: 0.8056 Discriminator Loss: 0.9487...

Epoch 1/10... Generator Loss: 0.8462 Discriminator Loss: 1.8311...
Epoch 1/10... Generator Loss: 2.6283 Discriminator Loss: 0.9064...
Epoch 1/10... Generator Loss: 2.0357 Discriminator Loss: 1.0606...
Epoch 1/10... Generator Loss: 0.8354 Discriminator Loss: 1.0690...
Epoch 1/10... Generator Loss: 2.5830 Discriminator Loss: 0.8270...
Epoch 1/10... Generator Loss: 1.0728 Discriminator Loss: 0.6448...
Epoch 1/10... Generator Loss: 1.1613 Discriminator Loss: 0.7259...
Epoch 1/10... Generator Loss: 0.4272 Discriminator Loss: 1.2481...
Epoch 1/10... Generator Loss: 0.7480 Discriminator Loss: 1.2353...
Epoch 1/10... Generator Loss: 0.9142 Discriminator Loss: 0.8230...

Epoch 1/10... Generator Loss: 1.6918 Discriminator Loss: 0.3787...
Epoch 1/10... Generator Loss: 1.6055 Discriminator Loss: 0.4696...
Epoch 1/10... Generator Loss: 1.1057 Discriminator Loss: 0.6439...
Epoch 1/10... Generator Loss: 2.9065 Discriminator Loss: 0.2758...
Epoch 1/10... Generator Loss: 0.8985 Discriminator Loss: 1.9248...
Epoch 1/10... Generator Loss: 0.4762 Discriminator Loss: 1.4877...
Epoch 1/10... Generator Loss: 0.5251 Discriminator Loss: 1.2103...
Epoch 1/10... Generator Loss: 0.4586 Discriminator Loss: 1.1997...
Epoch 1/10... Generator Loss: 0.9886 Discriminator Loss: 0.9923...
Epoch 1/10... Generator Loss: 2.0115 Discriminator Loss: 0.8776...

Epoch 1/10... Generator Loss: 1.6427 Discriminator Loss: 0.6676...
Epoch 1/10... Generator Loss: 0.8724 Discriminator Loss: 0.7769...
Epoch 1/10... Generator Loss: 1.7830 Discriminator Loss: 0.5226...
Epoch 1/10... Generator Loss: 2.6673 Discriminator Loss: 1.5352...
Epoch 1/10... Generator Loss: 1.5780 Discriminator Loss: 1.3724...
Epoch 1/10... Generator Loss: 1.9567 Discriminator Loss: 0.5973...
Epoch 1/10... Generator Loss: 0.5101 Discriminator Loss: 1.0826...
Epoch 1/10... Generator Loss: 1.4987 Discriminator Loss: 0.7464...
Epoch 1/10... Generator Loss: 0.4968 Discriminator Loss: 1.0991...
Epoch 1/10... Generator Loss: 0.8498 Discriminator Loss: 0.8096...

Epoch 1/10... Generator Loss: 3.0063 Discriminator Loss: 1.2510...
Epoch 1/10... Generator Loss: 3.0448 Discriminator Loss: 1.1804...
Epoch 1/10... Generator Loss: 3.5192 Discriminator Loss: 2.4012...
Epoch 1/10... Generator Loss: 1.7595 Discriminator Loss: 1.3114...
Epoch 1/10... Generator Loss: 1.6361 Discriminator Loss: 0.8126...
Epoch 1/10... Generator Loss: 1.9036 Discriminator Loss: 1.4999...
Epoch 1/10... Generator Loss: 0.9977 Discriminator Loss: 0.7851...
Epoch 1/10... Generator Loss: 0.5433 Discriminator Loss: 1.4005...
Epoch 1/10... Generator Loss: 0.4890 Discriminator Loss: 1.1589...
Epoch 1/10... Generator Loss: 0.2404 Discriminator Loss: 1.7234...

Epoch 1/10... Generator Loss: 0.1273 Discriminator Loss: 2.3453...
Epoch 1/10... Generator Loss: 0.8799 Discriminator Loss: 0.8954...
Epoch 1/10... Generator Loss: 0.4762 Discriminator Loss: 1.1700...
Epoch 1/10... Generator Loss: 0.4491 Discriminator Loss: 1.2321...
Epoch 1/10... Generator Loss: 0.5097 Discriminator Loss: 1.1084...
Epoch 1/10... Generator Loss: 0.9225 Discriminator Loss: 0.9852...
Epoch 1/10... Generator Loss: 2.4463 Discriminator Loss: 0.2251...
Epoch 1/10... Generator Loss: 1.3642 Discriminator Loss: 1.0487...
Epoch 1/10... Generator Loss: 2.3961 Discriminator Loss: 1.6950...
Epoch 1/10... Generator Loss: 1.1432 Discriminator Loss: 0.5795...

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.